The state estimation, sometimes referred to as wave filtering, improves the accuracy of the data by using the redundancy of the real-time measurement system, eliminates the wrong information caused by the random disturbance automatically, estimates or forecasts the operation state of the system. The function of the state estimation is to convert the raw data with low accuracy, imperfection and bad data into processed integral and reasonable data with little error. The chief task of the state estimation is to construct a network model using the telemetry information and remote signaling information of the real-time measurement system according to switch state, and to make raw-detection on real-time measurement data to eliminate the bad data and supplement the deficient measuring points with pseudo measurement to ensure the redundancy, thus estimating the system state, calculating the power flow distribution, meanwhile analyzing the operation state of the system using calculated results, enhancing the observability of the whole network.
It has been more than 30 years since the traditional method for state estimation was established by Fred Charles Schweppe of MIT in 1970. Different state estimators have different estimation criterions according to different objective functions used during the process of resolving the state variables. Presently, the estimation criterions used in power system comprises WLS (weighted least square) criterion, non-quadratic criterion, WLAV (weighted least absolute value), LMS (least median of squares) and LTS (least trimmed squares). The most widely-used criterion is the WLS estimation criterion, which has such advantages as simple model, small computational amount, and excellent statistical properties including optimal un-bias property on estimating measurement values of the ideal normal distribution. However, the disadvantage is its bad capability of robust, that is to say, the estimation result is easily affected by bad data and is far from the true value. Non-quadratic criterion, WLAV, LMS and LTS estimator are all robust estimators having notably high capability of robust, but such disadvantages as large computational amount and requiring to subjectively determined weighted factors are important factors that hinder the development of the robust estimators.
The present invention provides a method for state estimation in power system, which provides an estimation criterion with good capability of robust that can reflect the real distributing condition of the measurement error, constructs an optimal mathematical model based on the above estimation criterion with improved correctness of estimation by adding the injection power of link nodes into constraint conditions as the equality constraints, gives a principle for setting parameters in objective functions and a method for setting an initial value when resolving the said optimization problem.